9th International Conference on Quantitative InfraRed Thermography

July 2-5, 2008, Krakow - Poland

Ultrasound excited thermography of load bearing members used in constructional steelwork

by R. Plum* and T. Ummenhofer*

*Technische Universitaet Braunschweig - Institut fuer Bauwerkserhaltung und Tragwerk, Pockelsstr. 3, 38106 Braunschweig, Germany

Abstract

Ultrasonic lockin and burst-phase thermography establish more and more as fail-save NDE methods in the

scope of inspection of light-weight composite structures. The methods were not transferred to massive steel members used in constructional steelwork so far. Tests on a hot-rolled steel truss weighing more than 300 kg demonstrate that crack detection is possible but also show the known problems concerning infrared imaging of metal structures. Most critical points for the crack detection procedure are the strong dependency on the excitation frequency and the coupling between ultrasound converter and examined steel member. 1. Introduction

In zinc coated steel structures besides fatigue failures also damages due to liquid metal induced stress

corrosion cracking are to be detected [5, 10]. Standard NDE methods like magnetic particle inspection or dye penetrant testing are time-consuming if many details are to be examined. Radiographic tests are expensive, regular ultrasonic inspection and eddy-current testing are non imaging techniques. In the field of constructional steelwork there is no NDE method that is fast, fail-save, imaging and easy to handle. Thermography based approaches might be able to develop an innovative test method for crack recognition in massive steel components.

In non-destructive testing of structural elements ultrasound excited thermography offers various fields of application. The most practical experience with ultrasonic infrared imaging is available in testing of CFRP or similar types of composite materials. Fewer successful experiments have been conducted on massive metal components in the field of mechanical engineering like a cylinder head [14]. The idea of ultrasound excitation of structural members is to generate elastic waves that propagate inside the investigated structure. In case of internal failures like cracks or delaminations there is a chance to reach a state of oscillation in which boundary faces move relatively to each other. For steel members friction and plasticity have to be considered as the most relevant dissipative effects leading to heat generation [3]. With typical noise equivalent temperature differences (NETD) below 20 mK today’s cooled infrared camera systems easily detect smallest temperature variations on the component’s surface and offer new possibilities of nondestructive testing. Figure 1 shows the principle of a test setup for crack detection by means of ultrasound excited thermography.

Fig. 1. Principle of ultrasound excited thermography

2. Methods of ultrasound excited thermography In addition to the mono frequency excitation of a structure there are usually the following methods of

excitation and corresponding data processing available.

9th International Conference on Quantitative InfraRed Thermography

Ultrasonic lockin thermography If the excitation source at constant ultrasound frequency (e.g. 20 kHz) is being low-frequency amplitude-

modulated (e.g. sinusoidal, 1 Hz) the defect region dissipates energy corresponding to the temporal variation of the modulation. During excitation the component’s surface is recorded by a high-resolution infrared camera over several modulation periods. Finally a data processing is applied to the infrared sequence. For every pixel of the recorded stream the temperature amplitude and the phase shift related to the modulation history or to an arbitrary reference is calculated. This step can be done by means of a Fourier transform or by applying sine fittings to the data stream. Amplitude and phase information of all pixels are usually plotted as complete amplitude and phase images. High temperature amplitudes directly indicate regions of energy dissipation, phase images show heat sources and heat sinks and effectively suppress inhomogeneous surface emissivity and reflections.

Ultrasonic burst-phase thermography

The application of a burst excitation with a typical duration of a few milliseconds induces a large frequency

spectrum in the examined component. A Fourier-based evaluation at different frequencies allows the defect’s depth to be assessed as the thermal diffusion length is dependent on the analysed modulation frequency [15, 7, 12]. Burst-phase thermography is especially suited for detecting delaminations of CFRP components and in contrast to monofrequency excitation offers advantages concerning the sensitivity of defect detection [14].

Ultrasonic sweep thermography

A frequency sweep raising the ultrasonic frequency (e.g 20 - 25 kHz) is useful to realise the components

thermal response to ultrasonic excitation at various frequencies. During frequency sweep which can last seconds or minutes the oscillation amplitude is kept at a constant level. The procedure is appropriate for crack detection in metal components when there is a strong dependency on the ultrasound frequency. The preceding execution of an ultrasonic sweep can identify the excitation frequency leading to the highest temperature amplitudes reached in a lockin procedure. 3. Testing of massive steel components

A large number of conducted tests refer to structural elements made of composite materials like carbon

fibre reinforced plastic. The typical defect here is a localized delamination of the layered composition. The recognition of this type of defects is well documented in many cases [16, 11, 4]. With reference to [2] and [3] the maximum operating range of the ultrasound converter in order to obtain a fail-save defect detection amounts to 500 mm in CFRP components. During ultrasound excitation of CFRP components it has been reported of standing wave patterns. Beside structural imperfections that lead to locally increased energy dissipation the damping of the base material itself is sufficient to be detectable as dissipative heat generation patterns according to the excited vibration mode. Defect detection can be difficult or impossible if the failure’s position matches a nodal point of the oscillating structure. For that reason an additional frequency modulation during lockin or burst-phase thermography is recommended. The standing wave pattern can be superimposed by propagating waves suppressing the temperature patterns. In the scope of investigations on metal components the phenomenon only occurred while testing of thin sheets of steel [1]. Focussing on steel members with wall thicknesses greater than 8 mm no standing wave patterns were recognized during the conducted tests described below.

The defect recognition in CFRP samples was found to be marginal dependent on the ultrasonic frequency. A comparison of the thermal responses of CFRP and steel samples during an ultrasonic sweep thermography revealed that crack detection in metal components is strongly frequency dependent [6]. Agreeing to this insight defect selective heating of the examined steel components was only achieved at discrete restricted frequency bands. Investigations of turbine blades manifest that high infrared intensities are measured in the case of resonance only. However, not every resonant excitation leads to detection of all cracks [2, 9]. Contrary to the consequent recommendation to excite resonance frequencies only a multifrequency excitation leads to improved infrared contrast of the defect region during testing of thin aluminium samples [8].

It is assumed that thermographical crack detection in steel members is based on different mechanisms. On the one hand friction and clapping of the crack faces generates heat and on the other hand dissipative effects at the crack tip location occur. A numerical simulation of the ultrasonic excitation of a cracked aluminium plate indicates that plastic deformations at the crack tip have to be considered [3].

The following steel members have been examined using ultrasonic lockin thermography and sweep thermography.

Steel plate

Figure 2 shows a 600 mm long and 10 mm thick hot-galvanized steel plate which is cracked at the

geometrical notch due to fatigue tests. The crack length is estimated to be about 15 mm. The steel plate is equipped with reflective film in order to perform laser vibrometry measurements. To achieve a homogenous surface emissivity the notched area along the crack is coated with a graphit spray (ε ≈ 0,87 at room temperature). The coupling between ultrasonic converter and steel plate is done by means of a clamp or by stiff connection

NDE

using a mounting thread in the plate. For examination a BRANSON ultrasound generator with a frequency range of 15 to 25 kHz and adjustable oscillating amplitude is used. During ultrasonic excitation the crack can be clearly identified by means of a cooled FLIR Phoenix infrared camera at several ultrasound frequencies in the range of 20 - 23 kHz. Energy dissipation is to be recognized in top view along the crack faces as well as in thickness direction. In the used frequency range only one single ultrasound frequency was found that leads to heating at the crack tip location. It can be assumed that in this case cyclic plasticity at the crack tip is the prevailing effect. Figure 3 illustrates grey level infrared images in a colour scale at different excitation frequencies.

Fig. 2. Hot-galvanized steel plate with cracked notch

[1] iez

[4] eflec

[3] eflec

Fig. 3. IR images taken at different excitation frequencies (grey levels in colour scale) a) Visible crack in the top view, b) Visible crack along thickness direction, c) Visible crack tip

Fig. 4. Temperature response of the crack Fig. 5. Maximum temperatures occurred during frequency sweep during frequency sweep (back side)

In the course of all testing procedures it was found that crack detection is strongly dependent on the

chosen ultrasonic frequency. Figure 4 illustrates the characteristic thermal response of the defect region during a frequency sweep with 10 % of the maximum oscillating amplitude. On the back side of the steel plate two piezo ceramics are attached on both sides of the crack which were not used in the scope of these investigations. The maximum temperature for every pixel that appeared during the sweep is displayed in figure 5 (back side).

graphit coated area

Piezo elements

9th International Conference on Quantitative InfraRed Thermography

Obviously the shiny piezo element discs as an example for varying emissivity of the component’s surface prevents the crack from being easily detected in a reliable manner just by an ultrasonic sweep thermography.

Having found the most efficient ultrasound frequency for the examined component the following lockin excitation yields a temperature amplitude of about 0.85 K in the vicinity of the crack (Fig. 6). A plot of the relative phase shift clarifies the heat diffusion from the crack faces to the surroundings. Disturbing reflections due to varying surface emissivity are suppressed in the phase image. Figure 8 illustrates the temporal evolution of a single pixel representing the cracked area under sinusoidal modulated lockin excitation. Obviously the thermal equilibrium at the investigated location is not yet reached. The trend curve reveals an increase of the average temperature of almost 0.5 K during the conducted acquisition of 4.8 s.

Fig. 6. Temperature amplitude image

Fig. 7. Relative phase image Fig. 8. Temporal evolution of the temperature at crack location during lockin excitation (flockin ≈ 3.7 Hz)

Hot-rolled steel girder

The application of ultrasound excited thermography was transferred to massive load bearing members

used in constructional steelwork. The 3.2 m long hot-rolled steel girder IPE 500 shown in figure 9 has two quadratic web openings. Conducted 3-point bending fatigue tests caused crack initiation at the curved notches. The used ultrasound generation system enables to reach a noticeable state of vibration for the whole girder with a weight of almost 300 kg. As a result thermographical crack detection is possible as demonstrated in figure 10.

Fig. 9. Hot-rolled steel girder with web openings Fig. 10. Thermographical crack detection

NDE

Fig. 11. Thick-walled T-joint Fig. 12. Thermographical crack detection (grey levels in colour scale) a) monofreqency excitation; b) frequency modulated excitation

Welded thick-walled T-joint The third examined component is a part of a thick-walled welded beam with damages due to liquid metal

induced stress corrosion cracking. Figure 11 shows the T-joint with the vertical fillet weld cracked at many points along its height. An ultrasonic frequency sweep reveals that different frequencies lead to crack detection at different locations of the weld seam. One possibility to gather all defect regions during a single acquisition is to vary the ultrasound frequency fast enough so that heat diffusion at the first defect area lasts until the next defect is dissipating energy.

4. Relative motion of crack faces

The principal effects that lead to heat generation in cracks of metal components are not fully understood

so far. The influence of the vibration modes due to excitation on the occurring energy dissipating effects are to be investigated. Because of the strong frequency dependence of the crack detection it is assumed that a resonant excitation is required in the case of metal components. Furthermore crack detection is expected to correlate to the present crack. A finite element modal analysis of the steel plate proves that all three crack modes I-III can occur in the investigated frequency range.

By means of a fibre-optic interferometer used as differentially measuring vibrometer the out-of-plane oscillation of the crack faces was investigated at varying excitation frequency. Hence, two points on top of the steel plate at both sides of the crack were equipped with reflective stickers (compare fig. 3) to ensure a high resolution measurement of mode III movement of the crack faces. Figure 14 a) depicts the velocity amplitude vs. excitation frequency plot for the right crack face indicated in figure 3 c) in the range 20 - 23 kHz. Frequencies leading to crack detection are marked with a dot. The differential velocity amplitude of both measured points on the plate is shown in Figure 14 b).

Fig. 13. FE modal analysis of the steel plate shown in fig. 2 a) crack mode I at 22.87 kHz; b) crack mode II at 21.92 kHz; c) crack mode III at 20.28 kHz

9th International Conference on Quantitative InfraRed Thermography

Fig. 14. Velocity amplitude vs. excitation frequency plots; a) Velocity amplitude of the right measuring point in figure 2; b) Differential velocity amplitude of both measuring points in figure 2.

Comparing figure 14. a) and b) and focussing on the marked frequency 22.08 kHz which leads to crack face detection it can be recognized that the velocity amplitude of the single measurement is quite low. However, the differential velocity amplitude is very high and indicates a strong mode III movement of the crack faces. It can be assumed that in this case friction of the crack faces is the dominating effect leading to heat generation. Looking at the frequency 22.13 kHz it is notable that the measured point on the right side of the crack oscillates with a high velocity and thus with a large displacement amplitude. The differential velocity measured at 22.13 kHz demonstrates that both crack faces do not perform strong relative displacements according to mode III movement proved by rather low differential velocity amplitude. Nevertheless this excitation frequency was the only one to be found within the investigated range that enables to detect the crack tip shown in figure 3 c). From the results of the mode III vibrometer measurement it is assumed that at this frequency a regular bending mode shape occurs in which both crack faces move parallel to each other without typical mode III phase shift.

5. Summary

In contrast to carbon fibre reinforced polymer components with a typical thickness of a few mm which have

been often examined the success of crack detection in steel members is heavily dependent on the ultrasound excitation frequency. This is clarified by comparing thermo-acoustic spectra of samples made of CFRP and steel [13]. While the CFRP sample shows its failures almost within the whole excitation frequency range, the steel samples elevated thermal answer at crack locations can only be found at a few narrow frequency bands. It is assumed that in case of steel members only the excitation of eigenfrequencies leads to energy dissipation in cracked areas. The global eigenmode but especially the oscillating behaviour of the crack faces themselves is probably one of the most critical points for an efficient crack detection using ultrasound excited infrared thermography.

Three totally different steel components have been tested using a lockin and a sweep procedure. As a result ultrasound excited thermography has been attested the capability to develop to a fail-safe non-destructive test method that clearly visualizes cracks in steel members used in constructional steelwork. The reproducibility and the detection limits of the methods are to be investigated especially in the case of massive components. By means of laser vibrometry the correlation between vibration mode, ultrasonic frequency and energy dissipation in the crack area is assessable.

REFERENCES

[1] J. Baumann, R. Bilgram, V. Carl, T. Hierl, U. Netzelmann, H.R. Schubach, G. Zenzinger. Aktive thermographische Prüfmethoden zur Absicherung von Hochleistungsfertigungsverfahren - erste Ergebnisse. DGZfP Jahrestagung 2003, Mainz 26.-28.5.2003, V. 9, DGZfP-Berichtsband 83.

[2] J. Baumann, U. Netzelmann, R. Bilgram, T. Hierl, V. Carl, H.R. Schubach, G. Zenzinger. Untersuchungen zu aktiven thermographischen Prüfmethoden zur Absicherung von Hochleistungsfertigungsverfahren - Ergebnisse eines BMBF-Projekts. DGZfP-Jahrestagung 2005, Rostock 2.-4.5.2005, V. 57, DGZfP-Berichtsband 94.

[3] R. Bilgram, R. Stoessel, M. Diez-Vazquez, J. Robles Migueles, J. Schimitschek. Numerisches Modell zur Optimierung der Prüfbedingungen bei der US-angeregten Thermografie am Beispiel von Ermüdungsrissen und Delaminationen. DACH-Jahrestagung 2004 Salzburg 17.-19.5.2004, V35, DGZfP Berichtsband 89.

[4] L.D. Favro, X. Han, Z. Ouyang, G. Sun, H. Sui, R.L. Thomas. Infrared imaging of defects heated by a sonic pulse. Review of scientific instruments 71 (2000) 6 2418-2421.

[5] M. Feldmann, W. Bleck, P. Langenberg, T. Pinger, D. Tschickardt, A. Völling. Ermittlung der Rissanfälligkeit von Baustählen beim Stückverzinken. MP Materials Testing 49 (2007) 5 223-230.

22.13 kHz

22.13 kHz 22.08 kHz

22.08 kHz

NDE

[6] A. Gleiter, C. Spießberger, T. Zweschper, G. Busse. Improved ultrasound activated lockin-thermography using frequency analysis of material defects. International Conference on Quantitative Infrared Thermography (QIRT 2006), 27.-30.6.2006, Padova.

[7] A. Gleiter, G. Riegert, T Zweschper, G. Busse. Ultrasound-Lockin-Thermography for Advanced Depth Resolved Defect Selective Imaging. 9th European Conference on NDT, Berlin, September, 2006 ECNDT 2006 - We.3.8.2.

[8] X. Han, Z. Zeng, W. Li, S. Islam, J. Lu, V. Loggins, E. Yitamben, L.D. Favro, G. Newaz, R.L. Thomas. Acoustic chaos for enhanced detectability of cracks by sonic infrared imaging. Journal of applied physics 95 (2004) 7 3792-3797.

[9] C. Homma. Untersuchungen zu Mechanismus und technischer Umsetzung der akustischen Thermo-graphie. Dissertation, Universität des Saarlandes, 2007.

[10] W. Katzung, W.-D. Schulz. Beitrag zum Feuerverzinken von Stahlkonstruktionen - Ursachen und Lösungsvorschläge zum Problem der Rißbildung. Bericht Nr. 152 Gemeinschaftsausschuss Verzinken e.V., Sonderdruck aus Stahlbau 74 4; (2005).

[11] A. Mian, X. Han, S. Islam, G. Newaz. Fatigue damage detection in graphite/epoxy composites using sonic infrared imaging technique. Composites Science and Technology 64 (2004) 657-666.

[12] G. Riegert. Induktions-Lockin-Thermografie - ein neues Verfahren zu zerstörungsfreien Prüfung. Universität Stuttgart, Institut für Kunststofftechnik, Dissertation. 2007.

[13] C. Spießberger, A. Gleiter, G. Busse. Frequenzoptimierung ultraschallangeregter Thermografieverfahren. Thermografie-Kolloquium Stuttgart 2007.

[14] T. Zweschper, G. Riegert, A. Dillenz, G. Busse. Ultrasound burst phase thermography (ubp) for applications in the automotive industry. Quantitative Nondestructive Evaluation (QNDE) 2002.

[15] T. Zweschper, G. Riegert, A. Dillenz, G. Busse. Ultraschallangeregte Thermografie mittels frequenzmodulierter elastischer Wellen. DGZfP-Berichtsband 86, Thermografie-Kolloquium 2003.

[16] T. Zweschper, A. Dillenz, G. Riegert, D. Scherling, G. Busse. Ultrasound excited thermography using frequency modulated elastic waves. Insight 45 (2003) 3 178-182.